Safety Helmet Wearing Detection - IJSTMR...Jagruti Wani Dept. of Computer Late G. N. Sapkal College...
Transcript of Safety Helmet Wearing Detection - IJSTMR...Jagruti Wani Dept. of Computer Late G. N. Sapkal College...
Volume 3, Issue 06, June 2018 ISSN (online): 2456-0006
International Journal of Science Technology
Management and Research Available online at: www.ijstmr.com
IJSTMR © 2018 | All Rights Reserved 10
Safety Helmet Wearing Detection
Jagruti Wani Dept. of Computer
Late G. N. Sapkal College Of Engineering
Nashik ,M.H. India
Naina Nikam Dept. of Computer
Late G. N. Sapkal College Of Engineering
Nashik, M.H. India
Snehal Nandurkar Dept. of Computer
Late G. N. Sapkal College Of Engineering
Nashik, M.H. India
Sayali Sawale Dept. of Computer
Late G. N. Sapkal College Of Engineering
Nashik, M.H. India
M.G.Bhandari Asst.Prof.,Dept. of Computer
Late G. N. Sapkal College Of Engineering
Nashik, M.H. India
________________________________________________________________________________________________________
Abstract: Safety helmet wearing detection is very essential while traveling. We proposed a innovative and practical safety
helmet wearing detection method based on image processing and machine learning. At first, the extract background modeling
algorithm is exploited to detect motion object under a view of fix surveillant camera in power substation. After obtaining the
motion region of interest, the Histogram of Oriented Gradient (HOG) feature is extracted to describe inner human. And then,
based on the result of HOG feature extraction, the Support Vector Machine (SVM) is trained to classify pedestrians. Finally,
the safety helmet detection will be implemented by color feature recognition. We will also extract number plate of the user
without helmet which will be directly connected to their wallet with reference unique id. We will also add face detection feature
which will extract face and is send to the RTO office for further penalty.
Keywords: Histogram of oriented gradient, Support vector machine, Color feature recognition.
________________________________________________________________________________________________________
I. INTRODUCTION
Our aims to provide total safety for bike riders. Recently helmets have been made compulsory, but still people drive without
helmets. Pune City has approx. 35 lakh two-wheeler riders, which includes 500-600 accidents every year out of which 300-400 are fatal.
Pune ranks first in the city when it comes to two wheelers riders. In the last few years, there has been rapid increase in number of road
accidents. Due to rise in road accidents, it has now become necessary to generate a system to limit accidental deaths.
Lots of accidents occurred in power substation because of not wearing helmet while the working. Safety helmet wearing detection is a
very common and crucial task for surveillance in power substation.Whereas there are few researches for studying this problem by using
image processing techniques. Most researches focus on the approach investigating of motorcyclists whether wearing or not safety
helmets. Waranusast et al.developed an automatically detect system for motorcycle riders and was able to ascertain whether they are
wearing helmets or not. This system extracts the motion objects and trains a K-Nearest-Neighbor (KNN) classifier for detection. Silva et
al.exploited the Hough circular transformation to determine the shape of safety helmet and use the extracted Histogram of Oriented
Gradients (HOG) features to train a Multi-layer perceptron classifier, which can effectively and simply detect wearing helmet of
motorcyclists. In power substation, the surveillance camera is installed on the fixed location. So the view of camera is fixed which can
make sure that the background cannot change in frames. Consider this characteristic, we choose the ViBe background modelling
International Journal of Science Technology Management and Research
Volume 3, Issue 6, June 2018
www.ijstmr.com
IJSTMR © 2018 | All Rights Reserved 11
algorithm. Moreover, this method is fast and effective to determine the motion objects. In order to detect the people in power substation
whether wearing or not safety helmet, the second step is that obtaining the human location and image information. Thus, we extract the
HOG feature of people and train the SVM classifier for people to classify pedestrian in power substation. When we know the human
information in frames, we can utilize the color feature to detect safety helmet wearing situations.
II. PROBLEM DEFINITION
To find out by people or police department, lots of documentation and hard work is there also it takes the lot of time duration as well as
there is no guaranty of appropriate result. This application contains functionality to add complaint as well as view all complaints. By
using these complaints, Trust members will try to find lost person in various areas. This application will upload complaint on web server
which can be accessed by any of the trust member having this application.
III. PROPOSED SYSTEM
The main aim of this study is to propose and develop a system for automatic detection of helmets on public roads. The moto of this
project is accident prevention by using the methods of helmet authentication, fall detection etc.
Helmet Detection: More than one third who died in road accidents could have survived if they had worn a helmet. Studies shows that
usage of helmet can save accident death by 30 to 40%. The rate at which number of two wheelers in India is rising is 20 times the rate at
which human population is growing. The risk of death is 2.5 times more among riders not wearing a helmet compared with those wearing
a helmet.
SYSTEM ARCHITECTURE
Fig 1-: System Architecture
IV. Results Analysis
A machine learning technique, are used in identification of wearing helmets or not. A total of 100 images
belonging to 2 categories were used in this study. Classification processes with ANN was carried out for different
training-test rates of images in Table 1. 94% classification rate was achieved for the 70%-30% training-test
partition. According to the results, it was seen that texture features.
Experimental result in graph
International Journal of Science Technology Management and Research
Volume 3, Issue 6, June 2018
www.ijstmr.com
IJSTMR © 2018 | All Rights Reserved 12
Fig 2-: Classification accuracy
Table-1- summarized result
Table 1
CONCLUSION AND FUTURE SCOPE
An effort is made towards recognition of face and the obtained recognition accuracy is much. This method will be very beneficial
for finding missing person. This application will upload complaint on web server which can be accessed by any of the trust member
having this application. This project Finding Missing Person using Face Detection on Android Application presents the solution for this
problem. We are using four modules User, Police, Compliant holder, Admin for getting appropriate result.
Admin continuously Update database and Delete unnecessary data.
In future we intend to use more advanced safety measures like to check alcohol consumption, lane change detection, collision detection,
traffic information, e-toll collection, license renewal etc. We also think of applying deep neural network techniques & make
transportation more intelligent
REFERENCE
1. Jie Li, Huanming Liu, Tianzheng Wang, and Min Jiang” Safety Helmet Wearing Detection Based on Image Processing and Machine Learning” IEEE 2017.
2. R. Silva, K. Aires, and R. Veras, “Helmet detection on motorcyclists using image descriptors and classifiers,” 2014 27th SIBGRAPI Conference on Graphics Patterns
and Images, pp. 141-148, 2014.
3. W. Cai, J. Le, C. Jin, and K. Liu, Real-time image identification based anti-manmade misoperation system for substations, IEEE Trans. Power Del., vol. 2 no. 4, pp.
1748,1754, 2012.
4. Abu H. M. Rubaiyat, Tanjin T. Toma, Masoumeh Kalantari-Khandani” Automatic Detection of Helmet Uses for Construction Safety” IEEE 2016.
5. R.Waranusast, N. Bundon, V. Timtong, C. Tangnoi, and P. Pattanathaburt, Machine vision techniques for motorcycle safety helmet detection, 2013 28th International
Conference on Image and Vision Computing New Zealand, pp. 35-40, 2013.
6. Min Wang, lian-Qing Wang, Hong Qiao “Improved Shape Context Algorithm for Online Fast Recognition-An Application in Pedestrian Detection from a
Moving Vehicle”IEEE 2009.
Training Portion Classification Accuracy
30% - 70% 81.5
50% - 50% 89.9
70% - 30% 94
74
76
78
80
82
84
86
88
90
92
94
96
30% - 70% 50% - 50% 70% - 30%
Cla
ssif
icat
ion
A
ccu
racy
in
%
Training Portion
Classification Accuracy
Classification Accuracy
International Journal of Science Technology Management and Research
Volume 3, Issue 6, June 2018
www.ijstmr.com
IJSTMR © 2018 | All Rights Reserved 13
7. R. Wachal, J. Stoezel, M. Peckover and D. Godkin, ? Computer vision early-warning ice detection system for the Smart Grid,?in Transmission and Distribution
Conference and Exposition, pp. 1-6, 2012.
8. C. Fan, Y. Ni, R. Dou, D. Zhao, A. Zhao, and G. Huang, Modularization design for sequence control function in smart substation, Dianli Xitong Zidonghua (Automation
of Electric Power Systems), pp. 67-71, 2012.
9.O. Barnich, and M. Van Droogenbroeck, ViBe: a powerful random technique to estimate the background in video sequences, 2009 IEEE International Conference on
Acoustics, Speech and Signal Processing, pp. 945-948, 2009.
10. N. Dalal and B. Triggs, Histograms of oriented gradients for human detection, International Conference on Computer Vision and Pattern Recognition, pp. 886-
893,2005.